Compute & Storage · side by side
CoreWeave vs Railway
Compare CoreWeave, Railway by compute, bandwidth, storage, deployment limits, regions and support.
Compute & Storage pricing calculator
These providers bill in different units. Each estimate uses its own workload; there is no same-volume price winner.
CoreWeave
200
Railway
2
CoreWeave
Custom / not listed
Custom / unlisted usage pricing · 200 GPU hours
Railway
$5.00 / mo
Hobby · 2 vCPU-months
Listed USD plan fees and tracked usage only. Unlisted compute, transfer, extras, taxes and contract terms are excluded. Custom or unsupported plan usage is not estimated.
Compute & Storage overview & pricing
| Compute & Storage | CoreWeave | Railway |
|---|---|---|
| Provider | CoreWeave | Railway |
| Billing unit | GPU hours | vCPU-months |
| Plan at selected usage | Custom / unlisted usage pricing | Hobby |
| Monthly estimate | Custom / not listed | $5.00/mo |
| Workload | 200 GPU hours | 2 vCPU-months |
| Free access | Custom | Trial credits |
| Published plans | Not listed | Free, Hobby, Pro, Enterprise |
| Prices checked | Not listed | 2026-10-08 |
Provider
- CoreWeave
- CoreWeave
- Railway
- Railway
Billing unit
- CoreWeave
- GPU hours
- Railway
- vCPU-months
Plan at selected usage
- CoreWeave
- Custom / unlisted usage pricing
- Railway
- Hobby
Monthly estimate
- CoreWeave
- Custom / not listed
- Railway
- $5.00/mo
Workload
- CoreWeave
- 200 GPU hours
- Railway
- 2 vCPU-months
Free access
- CoreWeave
- Custom
- Railway
- Trial credits
Published plans
- CoreWeave
- Not listed
- Railway
- Free, Hobby, Pro, Enterprise
Prices checked
- CoreWeave
- Not listed
- Railway
- 2026-10-08
Compute & Storage monthly cost
CoreWeave: Not pricedRailway: $5.00
Different workloads, separate estimates
Separate application hosting, serverless requests and object storage. A GB of storage is not a GB of egress. Budget for compute duration, requests, builds and network transfer where charged, and check region and minimum-spend requirements.
Compute & Storage feature comparison
| Compute & Storage | CoreWeave | Railway |
|---|---|---|
| Projects | Not listed | 50 |
| Services per project | Not listed | 50 |
| CPU per service | Not listed | Up to 48 vCPU |
| RAM per service | Not listed | Up to 48 GB |
| Ephemeral disk | Not listed | 100 GB |
| Volume storage | Not listed | Up to 5 GB |
| Volumes per project | Not listed | 10 |
| Buckets per project | Not listed | 50 |
| Cron jobs per project | Not listed | 50 |
| Build image size | Not listed | 100 GB |
| Replicas per service | Not listed | Up to 6 |
| Running sandboxes per environment | Not listed | 50 |
| Default sandbox size | Not listed | 4 vCPU / 4 GB |
| Maximum sandbox size | Not listed | 8 vCPU / 8 GB |
| Idle timeout | Not listed | Up to 2 hours, or never |
| Project members | Not listed | 3 |
| Concurrent builds | Not listed | 3 |
| Build timeout | Not listed | 40 mins |
| IOPS | Not listed | 3,000 read/write operations per second |
| Class A operations | Not listed | Free |
| Class B operations | Not listed | Free |
| Bucket egress | Not listed | Free |
| Max storage | Not listed | 1 TB |
| Custom domains | Not listed | 2 |
| Service domains | Not listed | 4 |
| Log retention | Not listed | 7 days |
| Audit logs | Not listed | 48 hours |
Projects
- CoreWeave
- Not listed
- Railway
- 50
Services per project
- CoreWeave
- Not listed
- Railway
- 50
CPU per service
- CoreWeave
- Not listed
- Railway
- Up to 48 vCPU
RAM per service
- CoreWeave
- Not listed
- Railway
- Up to 48 GB
Ephemeral disk
- CoreWeave
- Not listed
- Railway
- 100 GB
Volume storage
- CoreWeave
- Not listed
- Railway
- Up to 5 GB
Volumes per project
- CoreWeave
- Not listed
- Railway
- 10
Buckets per project
- CoreWeave
- Not listed
- Railway
- 50
Cron jobs per project
- CoreWeave
- Not listed
- Railway
- 50
Build image size
- CoreWeave
- Not listed
- Railway
- 100 GB
Replicas per service
- CoreWeave
- Not listed
- Railway
- Up to 6
Running sandboxes per environment
- CoreWeave
- Not listed
- Railway
- 50
Default sandbox size
- CoreWeave
- Not listed
- Railway
- 4 vCPU / 4 GB
Maximum sandbox size
- CoreWeave
- Not listed
- Railway
- 8 vCPU / 8 GB
Idle timeout
- CoreWeave
- Not listed
- Railway
- Up to 2 hours, or never
Project members
- CoreWeave
- Not listed
- Railway
- 3
Concurrent builds
- CoreWeave
- Not listed
- Railway
- 3
Build timeout
- CoreWeave
- Not listed
- Railway
- 40 mins
IOPS
- CoreWeave
- Not listed
- Railway
- 3,000 read/write operations per second
Class A operations
- CoreWeave
- Not listed
- Railway
- Free
Class B operations
- CoreWeave
- Not listed
- Railway
- Free
Bucket egress
- CoreWeave
- Not listed
- Railway
- Free
Max storage
- CoreWeave
- Not listed
- Railway
- 1 TB
Custom domains
- CoreWeave
- Not listed
- Railway
- 2
Service domains
- CoreWeave
- Not listed
- Railway
- 4
Log retention
- CoreWeave
- Not listed
- Railway
- 7 days
Audit logs
- CoreWeave
- Not listed
- Railway
- 48 hours
Showing 15 of 27
Which compute & storage providers fit your needs?
Railway features & plan notes
- Usage-based
- Instant deploys
- Templates
- Includes $5 of monthly usage credits
- After credits are used, you'll only be charged for extra resource usage
- Up to 48 vCPU / 48 GB RAM per service
- Up to 5 replicas, at 8 vCPU / 8 GB RAM per replica
- Up to 5 GB storage
- Single developer workspace
- Community support
- 99.9% Availability Target
- 7-Day Log History
- Global regions
How to compare compute & storage providers
Separate application hosting, serverless requests and object storage. A GB of storage is not a GB of egress. Budget for compute duration, requests, builds and network transfer where charged, and check region and minimum-spend requirements.
Browse Compute & Storage pricingCoreWeave vs Railway — FAQ
CoreWeave: custom or unlisted usage pricing at 200 GPU hours. Railway: $5.00/month on Hobby at 2 vCPU-months. Different billing units are not an equal-workload comparison. Estimates exclude unlisted add-ons and taxes.
Not as one workload. Storage, bandwidth, requests and compute are separate billing units; estimates use independent controls when those units differ.
Build minutes, CPU time, memory, requests, outbound transfer, team seats and regional charges may be outside the tracked component.
H100 clusters, Kubernetes-native, Enterprise AI. Free access: Custom. Check plan-specific availability and current terms on the provider's official pricing page.
Usage-based, Instant deploys, Templates. Free access: Trial credits. Check plan-specific availability and current terms on the provider's official pricing page.